Why Does Surprisal From Larger Transformer-Based Language Models Provide a Poorer Fit to Human Reading Times?

27018
Back to Articles Dashboard

Knowledge Triples

Source Entity Relationship Target Entity
affects
influences